Dataset Open Access

# Experiments of the Paper "MORTY: A Toolbox for Mode Recognition and Tonic Identification"

Sertan Şentürk

### DataCite XML Export

<?xml version='1.0' encoding='utf-8'?>
<identifier identifierType="DOI">10.5281/zenodo.57999</identifier>
<creators>
<creator>
<creatorName>Sertan Şentürk</creatorName>
<affiliation>Universitat Pompeu Fabra</affiliation>
</creator>
</creators>
<titles>
<title>Experiments of the Paper "MORTY: A Toolbox for Mode Recognition and Tonic Identification"</title>
</titles>
<publisher>Zenodo</publisher>
<publicationYear>2016</publicationYear>
<subjects>
<subject>Ottoman-Turkish makam music</subject>
<subject>classification</subject>
<subject>mode recognition</subject>
<subject>tonic identification</subject>
<subject>k-nearest neighbors</subject>
<subject>pitch class distribution</subject>
<subject>toolbox</subject>
<subject>reproducibility</subject>
<subject>open source software</subject>
</subjects>
<dates>
<date dateType="Issued">2016-07-14</date>
</dates>
<resourceType resourceTypeGeneral="Dataset"/>
<alternateIdentifiers>
<alternateIdentifier alternateIdentifierType="url">https://zenodo.org/record/57999</alternateIdentifier>
</alternateIdentifiers>
<relatedIdentifiers>
<relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://zenodo.org/communities/ecfunded</relatedIdentifier>
<relatedIdentifier relatedIdentifierType="URL" relationType="IsPartOf">https://zenodo.org/communities/mir</relatedIdentifier>
</relatedIdentifiers>
<rightsList>
<rights rightsURI="info:eu-repo/semantics/openAccess">Open Access</rights>
</rightsList>
<descriptions>
<description descriptionType="Abstract">&lt;p&gt;This package contains the complete experimental data explained in:&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Karakurt, A., Şentürk S., &amp;amp; Serra X. (In Press).  MORTY: A Toolbox for Mode Recognition and Tonic Identification. 3rd International Digital Libraries for Musicology Workshop. &lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;Please cite the paper above, if you are using the data in your work.&lt;/p&gt;

&lt;p&gt;The zip file includes the folds, features, training and testing data, results and evaluation file. It is part of the experiments hosted in github (https://github.com/sertansenturk/makam_recognition_experiments/tree/dlfm2016) in the  folder call ".&lt;strong&gt;/data&lt;/strong&gt;". We host the experimental data in Zenodo (http://dx.doi.org/10.5281/zenodo.57999) separately due to the file size limitations in github.&lt;/p&gt;

&lt;p&gt;The files generated from audio recordings are labeled with 16 character long MusicBrainz IDs (in short "MBID"s) Please check http://musicbrainz.org/ for more information about the unique identifiers. The structure of the data in the zip file is explained below. In the paths given below &lt;em&gt;task&lt;/em&gt; is the computational task ("tonic," "mode" or "joint"), &lt;em&gt;training_type&lt;/em&gt; is either "single" (-distribution per mode) or "multi" (-distribution per mode),  &lt;em&gt;distribution&lt;/em&gt; is either "pcd" (pitch class distribution) or "pd" (pitch distribution), &lt;em&gt;bin_size&lt;/em&gt; is the bin size of the distribution in cents, &lt;em&gt;kernel_width&lt;/em&gt; is the standard deviation of the Gaussian kernel used in smoothing the distribution, &lt;em&gt;distance&lt;/em&gt; is either the distance or the dissimilarity metric, &lt;em&gt;num_neighbors&lt;/em&gt; is the number or neighbors checked in &lt;em&gt;k&lt;/em&gt;-nearest neighbor classification and &lt;em&gt;min_peak&lt;/em&gt; is the minimum peak ratio. 0 &lt;em&gt;kernel_width&lt;/em&gt; implies no smoothing. &lt;em&gt;min_peak &lt;/em&gt;always takes the value 0.15. For a thorough explanation please refer to the companion page (http://compmusic.upf.edu/node/319) and the paper itself.&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;folds.json: &lt;/strong&gt;Divides the test dataset (https://github.com/MTG/otmm_makam_recognition_dataset/releases) into training and testing sets according to stratified 10-fold scheme. The annotations are also distributed to sets accordingly. The file is generated by  the Jupyter notebook &lt;em&gt;setup_feature_training.ipynb (4th code block)&lt;/em&gt; in the github experiments repository (https://github.com/sertansenturk/makam_recognition_experiments/blob/master/setup_feature_training.ipynb).&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Features:  &lt;/strong&gt;The path is &lt;strong&gt;data/features/[distribution--bin_size--kernel_width]/[MBID--(hist &lt;/strong&gt;&lt;em&gt;or &lt;/em&gt;&lt;strong&gt;pdf)].json&lt;/strong&gt;. "pdf" stands for probability density function, which is used to obtain the multi-distribution models in the training step and "hist" stands for the histogram, which is used to obtain the single-distribution models in the training step. The features are extracted using the Jupyter notebook &lt;em&gt;setup_feature_training.ipynb (5th code block)&lt;/em&gt; in the github experiments repository (https://github.com/sertansenturk/makam_recognition_experiments/blob/master/setup_feature_training.ipynb)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Training: &lt;/strong&gt;The path is &lt;strong&gt;data/training/[training_type--distribution--bin_size--kernel_width]/fold(0:9).json]&lt;/strong&gt;. There are 10 folds in each folder, each of which stores the training model (file paths of the &lt;em&gt;distribution&lt;/em&gt;s in "multi" &lt;em&gt;training_type&lt;/em&gt; or the &lt;em&gt;distribution&lt;/em&gt;s itself in "single" &lt;em&gt;training_type&lt;/em&gt;) trained for the fold using the parameter set. The training files are generated by the Jupyter notebook &lt;em&gt;setup_feature_training.ipynb (6th code block)&lt;/em&gt; in the github experiments repository (https://github.com/sertansenturk/makam_recognition_experiments/blob/master/setup_feature_training.ipynb)&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Testing: &lt;/strong&gt;The path is &lt;strong&gt;data/testing/[task]/[training_type--distribution--bin_size--kernel_width--distance--num_neighbors--min_peak]&lt;/strong&gt;. Each path has the folders &lt;strong&gt;fold(0:9)&lt;/strong&gt;, which have the evaluation and the results files obtained from each fold. The path also has the &lt;strong&gt;overall_eval.json&lt;/strong&gt; file, which stores the overall evaluation of the experiment. The optimal value of &lt;em&gt;min_peak &lt;/em&gt;is selected in the 4th code block, testing is carried in the 6th code clock and the evaluation is done in the 7th code block in the Jupyter notebook &lt;em&gt;testing_evaluation.ipynb&lt;/em&gt; in the github experiments repository (https://github.com/sertansenturk/makam_recognition_experiments/blob/master/testing_evaluation.ipynb). &lt;br&gt;
&lt;strong&gt;data/testing/ &lt;/strong&gt;folder also contains a summary of all the experiments in the files &lt;strong&gt;data/testing/evaluation_overall.json &lt;/strong&gt;and &lt;strong&gt;data/testing/evaluation_perfold.json&lt;/strong&gt;. These files are created in MATLAB while running the statistical significance scripts. &lt;strong&gt;data/testing/evaluation_perfold.mat &lt;/strong&gt;is the same with the json file of the same filename, stored for fast reading.&lt;/li&gt;
&lt;/ul&gt;

</descriptions>
<fundingReferences>
<fundingReference>
<funderName>European Commission</funderName>
<funderIdentifier funderIdentifierType="Crossref Funder ID">10.13039/501100000780</funderIdentifier>
<awardNumber awardURI="info:eu-repo/grantAgreement/EC/FP7/267583/">267583</awardNumber>
<awardTitle>Computational models for the discovery of the world's music</awardTitle>
</fundingReference>
</fundingReferences>
</resource>

506
12
views